Real Time Object Tracking Algorithm

Abstract

A description of fast realization of the object tracking algorithm is presented. It is designed to track objects via a video camcorder mounted on an aircraft. The algorithm is based on a comparison of the correlation between the object image and regions of the current video frame together with analysis of inter-frame information. The video card programming technology CUDA has been used to provide real time computations on a standard PC. Also, the special robust version of the Kalman filter was developed to make the algorithm more stable. Tests showed applicability of the algorithm for solving tasks of the object tracking.